© 2007 Nature Publishing Group http://www.nature.com/naturebiotechnology
ARTICLES
Genome sequence of the lignocellulose-bioconverting
and xylose-fermenting yeast Pichia stipitis
Thomas W Jeffries1,2,8, Igor V Grigoriev3,8, Jane Grimwood4, José M Laplaza1,5, Andrea Aerts3, Asaf Salamov3,
Jeremy Schmutz4, Erika Lindquist3, Paramvir Dehal3, Harris Shapiro3, Yong-Su Jin6, Volkmar Passoth7 &
Paul M Richardson3
Xylose is a major constituent of plant lignocellulose, and its fermentation is important for the bioconversion of plant biomass
to fuels and chemicals. Pichia stipitis is a well-studied, native xylose-fermenting yeast. The mechanism and regulation of
xylose metabolism in P. stipitis have been characterized and genes from P. stipitis have been used to engineer xylose
metabolism in Saccharomyces cerevisiae. We have sequenced and assembled the complete genome of P. stipitis. The
sequence data have revealed unusual aspects of genome organization, numerous genes for bioconversion, a preliminary
insight into regulation of central metabolic pathways and several examples of colocalized genes with related functions.
The genome sequence provides insight into how P. stipitis regulates its redox balance while very efficiently fermenting
xylose under microaerobic conditions.
Xylose is a five-carbon sugar abundant in hardwoods and agricultural residues1, so its fermentation is essential for the economic
conversion of lignocellulose to ethanol2. Pichia stipitis Pignal (1967) is
a haploid, homothallic, hemiascomycetous yeast3,4 that has the
highest native capacity for xylose fermentation of any known
microbe5. Fed batch cultures of P. stipitis produce almost 50 g/l of
ethanol from xylose6 with yields of 0.35 to 0.44 g/g xylose (Fig. 1)7,
and they can ferment hydrolysates at 80% of the maximum
theoretical yield8.
P. stipitis Pignal (1967) is closely related to yeast endosymbionts of
passalid beetles9 that inhabit and degrade white-rotted hardwood10. It
forms yeast-like buds during exponential growth, hat-shaped spores
and pseudomycelia (Fig. 2), uses all of the major sugars found in
wood11 and transforms low-molecular weight lignin moieties12.
P. stipitis genes have been used to engineer xylose metabolism in
S. cerevisiae1, but regulation for ethanol production is problematic13.
S. cerevisiae regulates fermentation by sensing the presence of glucose,
whereas P. stipitis induces fermentative activity in response to oxygen
limitation14,15. Increasing the P. stipitis fermentation rate could greatly
improve its usefulness in commercial processes. Conversely, by using
knowledge of this native xylose-fermenting yeast, researchers could
improve xylose metabolism in S. cerevisiae.
We sequenced the P. stipitis genome to better understand its biology,
metabolism and regulation. In analyzing the genome we discovered
numerous genes for lignocellulose bioconversion (http://www.jgi.
doe.gov/pichia).
RESULTS
The 15.4-Mbp genome of P. stipitis was sequenced using a shotgun
approach and finished to high quality (o1 error in 100,000). The
eight chromosomes range in size from 3.5 to 0.97 Mbp, as previously
reported16. The finished chromosomes have only one gap in the
centromere region of chromosome 1. The Joint Genome Institute
(JGI) Annotation Pipeline predicted 5,841 genes (Table 1). A majority,
72%, have a single exon. Average gene density is 56%. Average gene,
transcript and protein lengths are 1.6 kb, 1.5 kb and 493 amino acids,
respectively. Expressed sequence tags (ESTs) support 40% of the
predicted genes with 84% showing strong similarity to proteins in
other fungi. Best bidirectional BLAST analysis of the gene models
against the Debaryomyces hansenii genome identified putative orthologs for 84% of the P. stipitis genes. Additionally, analysis of conservation between the genomes of P. stipitis and D. hansenii at the DNA
level using VISTA tools17 provided support for exons in 67.5% of the
P. stipitis genes.
Protein function can be tentatively assigned to about 70% of the
genes according to KOG (eukaryotic orthologous groups) classifications (Supplementary Fig. 1 online)18. Protein domains were predicted in 4,083 gene models. These include 1,712 distinct Pfam
domains. A PhIGs (phylogenetically inferred groups19, http://phigs.
org/) comparison of P. stipitis with eight other yeasts (Fig. 3)20
revealed 25 gene families representing 72 proteins specific to P. stipitis
(Supplementary Table 1 online). P. stipitis and D. hansenii share 151
gene families that are not found in the other genomes. The P. stipitis
1US Department of Agriculture, Forest Service, Forest Products Laboratory, One Gifford Pinchot Drive, Madison, Wisconsin 53705, USA. 2Department of Bacteriology,
University of Wisconsin-Madison, 420 Henry Mall, Madison, Wisconsin 53706, USA. 3DOE Joint Genome Institute, 2800 Mitchell Drive, Walnut Creek, California 94598,
USA. 4JGI/Stanford, Stanford Human Genome Center, 975 California Ave., Palo Alto, California 94304, USA. 5BioTechnology Development Center, Cargill, PO Box 5702,
Minneapolis, Minnesota 55440-5702,USA. 6Department of Food Science and Biotechnology Sungkyunkwan Univeristy, Suwon, Korea. 7Swedish University of Agricultural
Sciences (SLU), Dept. of Microbiology, Uppsala, Sweden. 8These authors contributed equally to this study. Correspondence should be addressed to T.W.J.
([email protected]) or I.V.G. ([email protected]).
Received 23 October 2006; accepted 22 January 2007; published online 4 March 2007; doi:10.1038/nbt1290
NATURE BIOTECHNOLOGY
VOLUME 25
NUMBER 3
MARCH 2007
319
70
35
60
30
50
25
40
20
30
15
20
10
10
5
0
0
0
1
2
Time (d)
3
(g/l)
40
Ethanol
45
80
Xylitol
90
Cell mass
© 2007 Nature Publishing Group http://www.nature.com/naturebiotechnology
Xylose (g/l)
ARTICLES
4
All of the genes for xylose assimilation, the oxidative pentose
phosphate pathway (PPP), glycolysis, the tricarboxylic acid cycle
(TCA) and ethanol production were present in isoforms similar to
those found in other yeasts (Fig. 5). Transcripts of GND1 are strongly
induced by growth on xylose under both aerobic and oxygen-limiting
conditions (Fig. 5). Transketolase (TKT1) is strongly induced on
xylose, and is one of the most abundant transcripts in the cell under
those conditions. Transcripts for PGI1, PFK1 and PFK2 were all
induced on xylose under oxygen limitation, but were low in number
under aerobic conditions (Fig. 5). Glyceraldehyde-3-phosphate dehydrogenase isoform 3 (TDH3), the gateway for glycolysis, was induced
by oxygen limitation on both glucose and xylose. Transcripts for PDC1
and ADH1 were low in number on xylose under oxygen-limited
Figure 1 Fermentation of xylose by Pichia stipitis CBS 6054 in minimal
medium. Xylose, blue; ethanol, green; cell mass, red; xylitol, gold.
gene set was missing 81 gene families (442 proteins) relative to the
other yeast genomes in the analysis.
The most frequent domains include protein kinases, helicases,
transporters (sugar and MFS) and domains involved in transcriptional
regulation (fungal specific transcription factors, RNA recognition
motifs and WD40 domains). A majority is shared with other
hemiascomycota. These range from 1,534 common with
Schizosaccharomyces pombe to 1,639 with D. hansenii. One of the few
P. stipitis–specific domains (Supplementary Table 2 online) belongs to
glycosyl hydrolase Family 10, a subgroup of cellulases and xylanases.
Several gene families expanded in P. stipitis show some sequence
similarity to hyphally regulated cell wall proteins, cell surface flocculins,
agglutinin-like proteins and cytochrome p450 nonspecific monooxygenases. Members of these expanded families are poorly conserved
and often occur near chromosome termini (within 35,000 bp).
Chromosomal segments that retain the ancestral gene groupings can
be identified between P. stipitis and D. hansenii. A total of 263 orthology
segments were found, encompassing 4,456 of the genes (10,950,900 bp)
in the P. stipitis genome, and 4,689 genes (9,057,788 bp) in the
D. hansenii genome. On average, each block in the P. stipitis genome
encompasses 16.9 genes and is 41.6 kb in length. The largest of these
orthologous chromosomal segments, which is 301.9 kb in length and
encompasses 125 genes, is between P. stipitis chromosome
6 and D. hansenii chromosome F (Fig. 4). The rates of genomic
rearrangement observed here are consistent with previously reported
rates between D. hansenii and Candida albicans21.
P. stipitis uses the alternative yeast nuclear codon (12) that substitutes serine for leucine when CUG is specified22. A count of CUG
usage showed 15,265 occurrences in 4,238 open reading frames
(ORFs), or about 72% of all gene models. Nine out of the 21 ORFs
having 18 or more CUGs in the gene model occurred at or near a
terminus of chromosomes 4, 8, 7 or 1. All gene models having a large
number of CUGs in the ORF were large (42,500 bp), very large
(45,000 bp), repetitive, hypothetical or poorly defined.
P. stipitis possesses genes for a number of transporters that are
similar to putative xylose transporters from Debaromyces hansenii
(NCBI AAR06925) and Candida intermedia (GXF1, EMBL AJ937350;
GXS1, EMBL AJ875406)23. C. intermedia GXF1 has the closest
similarity to the previously described, closely related SUT1, SUT2
and SUT3 genes of P. stipitis24 and to the P. stipitis SUT4 gene, which
was identified in the present genome sequence (Supplementary Fig. 2
online). Notably, SUT2 and SUT3 are each located very near the ends
of their respective chromosomes. Available EST data do not show
their expression.
320
1 µm
1 µm
Figure 2 Morphology under various conditions. Pichia stipitis growing
exponentially with bud scars (top); P. stipitis hat-shaped spores seen
from top and side (center); Pseudomycelia formed under carbon-limited
continuous culture (bottom). Photo by Thomas Kuster, USDA, Forest
Products Laboratory.
VOLUME 25
NUMBER 3
MARCH 2007
NATURE BIOTECHNOLOGY
ARTICLES
Table 1 General characteristics of several yeast genomes
Genome Size
(Mb)
Avg. G+C content (%)
Total CDS
Avg. gene
density (%)
Avg. G+C in CDS (%)
Avg. CDS size
(codons)
Maximum CDS size
(codons)
Source
P. stipitis
S. cerevisiae
15.4
12.1
41.1
38.3
5,841
5,807
55.9
70.3
42.7
39.6
493
485
4,980
4,911
JGI
Dujon20
C. glabrata
K. lactis
12.3
10.6
38.8
38.7
5,283
5,329
65.0
71.6
41.0
40.1
493
461
4,881
4,916
Dujon20
Dujon20
D. hansenii
Y. lipolytica
12.2
20.5
36.3
49.0
6,906
6,703
79.2
46.3
37.5
52.9
389
476
4,190
6,539
Dujon20
Dujon20
© 2007 Nature Publishing Group http://www.nature.com/naturebiotechnology
Species
conditions. The five NADP(H)-coupled alcohol dehydrogenases
(ADH3, 4, 5, 6 and 7) could maintain cofactor balance between
NADH and NADPH. Transcripts for mitochondrial isocitrate dehydrogenases (IDH1, IDH2) are elevated on xylose under oxygen-limited
conditions, as are those for malate dehydrogenase (MDH1), fumarase
(FUM1) and succinic dehydrogenase (SDH1). The transcript for
2-ketoglutarate dehydrogenase (KGD1), which generates NADH in
the TCA cycle, was reduced during cultivation on xylose.
An NAD-specific glutamate dehydrogenase (GDH2), a glutamate
decarboxylase (GAD2), and two NADP-dependent succinate semialdehyde dehydrogenases (UGA2, UGA22) constitute a bypass to convert a-ketoglutarate into succinate and NADH into NADPH when
cells are growing on xylose. The NADH-specific GDH2 is elevated on
xylose under oxygen limitation, whereas the NADPH-linked glutamate dehydrogenase 3 (GDH3) is not. The increased level of GDH2
could also account for the decreased level of KGD2 when cells are
growing on xylose. Distinctly different sets of genes are strongly
induced under oxygen-limited growth on glucose and xylose (Supplementary Table 3 online). On xylose, the transcript for fatty acid
synthase 2 (FAS2) and the stearoyl-CoA desaturase, (OLE1) are
strongly induced under oxygen limitation.
A Family 10 xylanase, XYN1, was found along with several Family 5
endo-1,4-b-glucanases or cellodextrinases (EGC1, EGC2 and EGC3).
EGC2 is strongly expressed in cells growing on xylose. The three exo1,3-b-glucosidases (EXG1, EXG2, EXG3) could help the beetle host
digest fungal hyphae. The Family 17 soluble cell wall glucosidases
(SCW4.1, SCW4.2 and SCW11) along with the Family 17 exo-1,3-bglucanases (BGL2, BOT2), are most likely involved in cell wall
expansion and growth. Family 3 b-glucosidases (BGL1-7) can have
5
94
21
473
5
90
35
383
30
151
~1,000
25
516
33
121
Eremothecium gossypii
Kluyveromyces lactis
Candida glabrata
Saccharomyces cerevisiae
Debaryomyces hansenii
Pichia stipitis
Candida lusitaniae
Yarrowia lipolytica
Schizosaccharomyces pombe
Figure 3 Phylogenetic tree of seven sequenced hemiascomycetous yeast
genomes based on multiple alignment of 94 single-copy genes conserved
in 26 taxonomic groups (see Methods). Numbers next to each branch
correspond to the number of families (clusters) specific to a genome or
a group of genomes leading to this node.
NATURE BIOTECHNOLOGY
VOLUME 25
NUMBER 3
MARCH 2007
activity against cellobiose or xylobiose. Of the seven found in P. stipitis,
BGL4 is most similar to classical cellobioses and BGL7 is expressed
most when cells are growing on xylose (Supplementary Table 3). The
Family 2 b-mannosidases (BMS1, MAN2) are probably responsible for
the capacity of this yeast to grow on and ferment mannan oligosaccharides, but the endo-1,6-a-mannosidases (DCW1, DFG5) are
likely involved in yeast cell wall expansion during growth, since they
are present when cells are growing on either glucose or xylose.
P. stipitis has four a-glucosidases (MAL6-9) and a Family 31
a-glucosidase/a-xylosidase (YIC1). Of these, only MAL8 was detected
when cells were grown on xylose.
Five salicylate hydroxylases (NHG1.1, NHG 1.2, NHG2, NHG3,
NHG4) are scattered throughout the genome. Only NHG2 shows
conservation relative to D. hansenii. The rest of the genes and their
surrounding loci have no identity to proteins found in other yeasts.
The genome contains almost 60 ORFs that are identified as chitinases
according to KOG classification. Only four (CHT1, CHT2, CHT3,
CHT4) are likely to be involved in degradation of insect or fungal cell
walls. The remaining models are mucin-like proteins. MUC1 appears
four times in nearly identical copies, and segments exist in B25
copies, suggesting expansion through frequent duplication.
A gene for DUR1 (DUR1,2, urea amidolyase) is immediately
adjacent to the urea transporter DUR3.1. In addition to DUR3.1
P. stipitis has two other genes for urea transport. DUR3.2 and DUR5.1.
These are not found in any of the other yeasts with sequenced
genomes. Multiple copies of similar transporters (e.g., DUR4,
DUR5.2, DUR5.3, DUR8) are also found in P. stipitis.
b-glucosidases were often adjacent or proximal to genes with related
functions (Supplementary Table 4 online). For example, on either
side of the b-1,4 endoglucanase EGC2, one finds BGL5 and the
probable hexose transporter, HXT2.4. EGC3 is adjacent to HXT2.1.
BGL6 is adjacent to EGC1, and BGL3 is adjacent to SUT3, and BGL1
and HXT2.6 are adjacent to SUT2. Both of the putative P. stipitis
b-mannosidases (BMS1, MAN2) are adjacent or proximal to putative
sugar permeases (LAC3 and LAC2, respectively).
One of the most conspicuous examples of tandem genes with
related functions is the MAL3 locus (Fig. 6). This site contains a
maltose permease, MAL3, and the a-glucosidase, AGL1. Adjacent to
MAL3 is MAL5, which is adjacent to YIC1, an a-glucosidase. Flanking
this complex are a fungal transcriptional regulatory protein, SUC1.2,
similar to MAL-activator proteins25, and a second putative fungalspecific regulatory protein, SUC1.4. Elsewhere in the genome, on
chromosome 6, the a-glucosidase, MAL8, is immediately adjacent to
the maltose permease, MAL4.
We identified a number of transposable elements using a composite
library of fungal repeats26. The most abundant include long terminal
repeat retrotransposons Tdh5, Tdh2, Tse5, pCal, most of which
are present in D. hansenii27. Single copies of DNA mediated elements
Ty1-I, Mariner-5 and Folyt1 were reported earlier in fungi28. Copies
321
ARTICLES
Pichia stipitis chromosomes
© 2007 Nature Publishing Group http://www.nature.com/naturebiotechnology
Chr
1.1
Chr
1.2
Chr
2
Chr
3
Chr
4
Chr
5
Chr
6
Chr
7
Chr
8
Debaryomyces hansenii
Chromosomes colors
Chr A
Chr B
Chr C
Chr D
Chr E
Chr F
Chr G
Figure 4 Orthologous chromosomal segments observed between Pichia
stipitis and Debaryomyces hansenii.
of the retrotransposon Tps5 show one well-defined locus on
each chromosome.
DISCUSSION
The CBS 6054 strain was isolated from insect larvae, and other yeast
strains closely related to P. stipitis have been isolated from the guts of
wood-inhabiting passalid beetles10, suggesting that this family of yeasts
has evolved to inhabit an oxygen-limited environment rich in partially
digested wood. The presence of numerous genes for endoglucanases
and b-glucosidases, along with xylanase, mannanase and chitinase
activities indicates that it could metabolize polysaccharides in the
beetle gut. Various strains of P. stipitis have been reported to ferment
322
cellobiose to ethanol29. Exo-1,4-cellobiohydrolases, which are responsible in part for the degradation of cellulose, produce cellobiose from
cellulose and most endo-1,4-xylanases produce a mixture of xylose,
xylobiose and xylotriose. b-glucosidases and b-xylosidase activities are
therefore very useful traits when cellulose and hemicellulose saccharification is combined with fermentation.
Genes for xylose assimilation (XYL1, XYL2) were not expressed in
the presence of glucose in the medium. GND1 and TKT1 were
substantially elevated when growing on xylose, which reflects the
increased activity of the PPP for xylose metabolism. PGI1, PFK1
and PFK2 were elevated most with cells growing on xylose under
oxygen-limited conditions. Presumably elevated PGI1 is necessary to
cycle fructose 6-phosphate (F6P) through the oxidative PPP whereas
PFK1 and PFK2 take F6P into glycolysis. GLK1 was elevated in
cells growing on xylose aerobically, which could reflect carbon
catabolite de-repression.
Excess NADH is generated during growth on xylose1, which
necessitates cofactor regeneration. KGD2, which forms NADH in the
TCA cycle, was three times higher in cells growing on glucose over
those on xylose. Gdh2 consumes NADH while generating NAD+, and
leads into a pathway that eventually consumes NADH while generating NADPH. A similar pathway was previously engineered in
S. cerevisiae to reduce cofactor imbalances when cells are growing
on xylose30, but it appears to exist naturally in P. stipitis.
P. stipitis has a complete mitochondrial respiration system including
a SHAM-sensitive terminal alternative oxidase (AOX1 or STO1)31, and
NADH dehydrogenase Complex I, both of which are lacking in
S. cerevisiae. Without Complex I, S. cerevisiae has less capacity for
ATP generation through oxidative phosphorylation. AOX1 is not
proton translocating, but it could scavenge for oxygen to balance
cofactors. The EST data do not provide evidence of a role for AOX1
in xylose fermentation.
The abundance of genes for NADP(H) oxidoreductase reactions
suggests that P. stipitis is capable of various strategies for balancing
NAD and NADP-specific cofactors. Not least among these is FAS2,
which appears to be highly active when cells are growing under
oxygen-limited conditions on xylose. Fas2 synthesizes long chain
acyl-CoA precursors of fatty acids that could serve as a reductant
sink. Transcripts for fatty acid synthesis including OLE1 and, particularly, FAS2 were elevated in oxygen-limited, xylose-grown cells,
indicating that reductant is channeled into lipid synthesis under
oxygen limitation.
Colocation of genes having different but related functions (e.g., a
permease with a hydrolase for maltose) occurs with high frequency in
P. stipitis, but it is not confined to this yeast. For example, the
association between urea permease and urea amidolyase is found
throughout the sequenced yeast genomes. DUR1,2 is immediately
downstream of an ortholog of DUR3 in a wide variety of ascomycetous yeasts including D. hansenii, Candida glabrata, Kluyveromyces
lactis, S. cerevisiae and Yarrowia lipolytica, and there are several
examples of the MAL3 locus in other yeasts. Other proximal associations, such as those between BGL and SUT genes, appear to be unique
to P. stipitis.
Proximal orthologs with strong similarity to EGC2
(DEHA0G07095g),
BGL5
(DEHA0G07183g)
and
HXT2.4,
(DHEA0G07117 and DHEA0g07139) are also found in D. hansenii,
but their locations and arrangement are different in that yeast, so
although the orientation and number of these genes change, functional relationships remain. The closest sequenced relative to P. stipitis,
D. hansenii, does not have similar correlations between BGL and SUT
genes even though it possesses six putative b-glucosidases and three
VOLUME 25
NUMBER 3
MARCH 2007
NATURE BIOTECHNOLOGY
ARTICLES
P. stipitis, some genes at chromosome termini
have orthologs proximal to functionally
2
related genes at sites deeper within the chro5
Glucose
1
mosomes, so a similar mechanism could be
HXK1
0
0
GLK1
Xylose
ZWF1
GND1
working here.
HXK1
GLK1
ZWF1
GND1
XYL1
Genes in telomeric regions might be under
G-6-P
Ru-5-P
7
20
RPE1
6
Xylitol
less
selective pressure because of silencing. In
15
PGI1
5
RKI1 XYL2
4
S. cerevisiae the COMPASS histone methyl10
F-6-P
Xylulose
3
5
transferase carries out telomeric silencing of
2
TKT1
XYL3
PFK1
1
0
gene expression33. The P. stipitis genome conPFK2
X-5-P
0
XYL1 XYL2 XYL3
PGI1 PFK1 PFK2
E-4-P
tains a COMPASS homolog (SET1), so the
Ri-5-P
F-1,6-BP
12
TKT1
TAL1
10
same mechanism might be functioning here.
FBA1
40
8
35
6
Ga-3P
Without selective pressure, genes in the teloTPI1
30
4
S-7-P
Ga-3P
DHAP
25
2
meric regions could diverge more rapidly. We
20
0
25
15
TDH1,2,3
TPI1
FBA1
10
noted that genes occurring at chromosome
20
5
15
25
0
Gly-1,3P
termini often had a high frequency of CUG
10
20
TAL1
TKT1
5
15
usage, which might indicate genetic drift.
PGK1
10
0
TDH3 TDH2 TDH1
5
The proximal location of glucosidases to
0
3-PG
PGK1 GPM1.1 GPM2
sugar transporters and adjacency of urea amiGPM1.1
6
GA
XA
GPM2
5
dolyase to urea permease suggest that these
4
3
2-PG
GOL
XOL
loci might be coregulated. In S. cerevisiae,
30
2
25
1
ENO1
20
genes
for a-glucosidase and maltose per0
15
ENO1
PYK1
10
mease are adjacent. Each complete MAL
PEP
5
8
0
PCK1
6
locus consists of maltose permease, maltase
PDC1 PDC2 ADH1
PYK1
4
0.8
ADH1
PDC1
2
PYC1
and a transcription activator. The MAL
Acetal0.6
Ethanol
PYR
0
dehyde
0.4
PYC1
PCK1
loci each map to the telomeric region of a
LAT1
PDA1
0.2
ALD1
different chromosome34.
PDB1
0.0
4
ALD1
ACS1
ACS1
20
In eukaryotes, coregulated genes distal
3
AcCoA
Acetate 16
3
CO2 4
2
12
from
one another can be physically coloca3
2
1
CIT1
8
2
0
1
4
1
lized
in nuclear ‘transcriptional factories’.
PDA1 PDB1 LAT1
0
0
0
Isoct ACO1
CIT1
IDH1
IDH2
ACO1
Osborne et al. proposed that linked genes
IDH1,2
14
ICL1
12
3
are more likely to occupy a transcriptional
Oaa
GDH2,3
10
2
8
+
NAD(P)H
factory than genes in trans. In the human
6
NAD(P)
Glo
MDH1
1
L-Glu 4
KGD2
MLS1.2
2
transcriptional map, genes with increased
0
+
0
Mal
NADPH
MDH1 MDH2
Succ
GDH2
GDH3
GAD2
expression occur in gene-dense regions35.
+
+
(NAD ) (NADP )
UGA2,22
FUM1
SDH1,2,4,6
CO2
Adjacent
eukaryotic genes are more frequently
1.5
Fum
+
4-AB
NADP
1.0
coexpressed
than is expected by chance and
8
0.5
SUC-SA
UGA1
0.0
6
8
coexpressed
neighboring
genes are often funcFUM1 MLS1.2
4
6
4
tionally related. For example, in Arabidopsis
2
2
0
thaliana, 10% of the genes occur in 266
0
KGD2
SDH1 SDH2 SDH4 SDH6
groups of large, coexpressed, chromosomal
regions distributed throughout the genome36.
Figure 5 Relative abundances of transcripts in the central metabolic pathways of Pichia stipitis.
Cells were grown batch-wise on minimal defined medium under four conditions: glucose aerobic (GA),
One published model37 encapsulates the
xylose aerobic (XA), glucose oxygen limited (GOL) and xylose oxygen-limited (XOL). cDNA was
advantages of proximal colocation of actively
harvested and sequenced.
transcribed genes: the concentration of RNA
polymerase II is 1,000-fold higher in a tranSUT genes very similar to those found in P. stipitis, which suggests scription factory than in the whole nucleus; modifications occurring
that these proximal relationships evolved through selective pressure during transcription leave the promoter open to new transcript
in P. stipitis. Two out of the six genes of the MAL3 locus appear to initiation; after being released at the termination, promoters in the
be conserved in C. albicans, and four out of the six are conserved vicinity of a transcription factory are more likely to encounter
in D. hansenii.
Members of multigene families are found near S. cerevisiae telo25,000
30,000
35,000
40,000
45,000
meres and are repeated elsewhere in the genome. It has been proposed
that the concentration of multigene families in the telomere-adjacent
Vista conservation in D. hansenii
regions reflects a recombination-mediated dispersal mechanism32. In
3
15
© 2007 Nature Publishing Group http://www.nature.com/naturebiotechnology
10
Vista conservation in C. albicans
Figure 6 The MAL3 locus of P. stipitis. Two putative a-glucosidases (YIC1,
AGL1) and two putative maltose permeases (MAL3, MAL5) are colocated
along with two putative fungal transcriptional regulators (SUC1.2, SUC1.4)
within 16 kbp on chromosome 6.
NATURE BIOTECHNOLOGY
VOLUME 25
NUMBER 3
MARCH 2007
SUC1.4
YIC1
MAL5
MAL3
AGL1
SUC1.2
Genes in the P. stipitis locus
323
© 2007 Nature Publishing Group http://www.nature.com/naturebiotechnology
ARTICLES
machinery for transcriptional initiation again. These factors would all
favor survival of strains in which genes with related function and
regulatory features would be colocated in the genome.
The P. stipitis genome is endowed with numerous genes and
physiological features enabling it to ferment a wide variety of sugars
derived from lignocellulose, including a high capacity for using
cellobiose and other oligomers. We discerned structural features
such as genes with related functions proximal to one another that
suggest the combined gene activities enhance survival. The genes can
occur separately, but proximal location could affect their mutual
function and the probability of co-inheritance. If some gene families
persist in multiple copies simply from the advantage of higher
transcript levels, then evolution toward higher promoter strength
would be sufficient. Their presence in multiple copies suggests multiple functions. If chromosomal colocation affects expression, this
would have implications with respect to the design and placement
of genes for metabolic pathway engineering.
METHODS
Yeast strain. Pichia stipitis Pignal (1967), synonym Yamadazyma stipitis
(Pignal) Bilon-Grand (1989), (NRRL Y-11545 ¼ ATCC 58785 ¼ CBS 6054 ¼
IFO 10063) was obtained as a lyophilized powder. It was revived and streaked
on yeast extract, peptone, dextrose (YPD) agar to obtain isolated colonies. A
single colony was transferred to 150 ml of YPD broth. To test for contamination, the overnight culture was observed under the microscope and streaked in
both YPD and LB plates. For fermentation studies, cells were grown in 125-ml
Erlenmeyer flasks containing 50 ml of 1.67 g/l yeast nitrogen base (YNB) with
2.27 g/l urea and 80 g/l xylose. The YNB and urea solutions were filter sterilized
in a 20 solution and added to the sugar, which was sterilized separately by
autoclaving. For mRNA preparation, cells were grown in YPD, which was
prepared as described38 except that sugars were autoclaved separately from
the basal medium. Yeast, peptone, xylose (YPX) was similar to YPD but
replaced dextrose with xylose. Preparation of mRNA was by the method
previously described22.
DNA preparation. Yeast genomic DNA was prepared following a published
protocol39. Two extra phenol:chloroform/chloroform extractions and ethanol
precipitation were carried out. To prevent shredding of the DNA, the sample
was not vortexed. The final gDNA concentration was 500 ng/ml as determined
by optical density at 260 nm.
cDNA library construction and sequencing. P. stipitis CBS 6054 was grown at
30 1C in 200 ml of either YPD or YPX in either a 2.8 l flask shaken at 300 r.p.m.
or a 500 ml flask shaken at 50 r.p.m. Aerobic cultures were inoculated with a
low cell density (0.025 mg/ml), shaken at 200 r.p.m. and harvested at a cell
density of less than 0.5 mg/ml. Oxygen-limited cultures were inoculated with a
high cell density (2.5 mg/ml), shaken at 100 r.p.m. and harvested at 5 mg/ml.
Cells were collected by centrifugation at 4 1C and 9,279g. Cells were suspended
in water and centrifuged at 835g for 5 min. Cells were then frozen in liquid N2.
Poly A+ RNA was isolated from total RNA for all four P. stipitis samples using
the Absolutely mRNA Purification kit (Stratagene). cDNA synthesis and
cloning was a modified procedure based on the SuperScript plasmid system
with Gateway technology for cDNA synthesis and cloning (Invitrogen). We
used 1–2 mg of poly A+ RNA, reverse transcriptase SuperScript II (Invitrogen)
and oligo dT primer (5¢-GACTAGTTCTA GATCGCGAGCGGCCGCCC
TTTTTTTTTTTTTTT-3¢) to synthesize first-strand cDNA. Second-strand
synthesis was performed with Escherichia coli DNA ligase, polymerase I and
RNaseH followed by end repair using T4 DNA polymerase. The SalI adaptor
(5¢-TCGACC CACGCGTCCG and 5¢-CGGACGCGTGGG) was ligated to the
cDNA, digested with NotI (NEB), and subsequently size selected by gel
electrophoresis (1.1% agarose). Size ranges of cDNA were cut out of the gel
(Low insert size: 600–1.2 kb; Medium insert size: 1.2 kb–2 kb; High insert size:
42 kb) and directionally ligated into the SalI- and NotI-digested vector
pCMVsport6 (Invitrogen). ElectroMAX T1 DH10B cells were transformed by
the ligation (Invitrogen).
324
Library quality was first assessed by PCR amplification of the cDNA inserts
of 20 clones with the primers M13-F (5¢-GTAAAACGACGGCCAGT-3¢) and
M13-R (5¢-AGGAAACAGCTATGACCAT-3¢) to determine insert rate. Clones
for each library were inoculated into 384-well plates (Nunc) and grown in LB
for 18 h at 37 1C. DNA template for each clone was prepared by rolling circle
amplification and sequenced using primers (FW: 5¢-ATTTAGGTGACACTA
TAGAA-3¢ and RV 5¢-TAATACGACTCACTATAGGG-3¢), using Big Dye
chemistry (Applied Biosystems). The average read length and pass rate were
753 (Q20 bases) and 96%, respectively.
EST sequence processing and assembly. The JGI EST Pipeline begins with the
cleanup of DNA sequences derived from the 5¢and 3¢ end reads from a library
of cDNA clones. The Phred software40 is used to call the bases and generate
quality scores. Vector, linker, adaptor, poly-A/T and other artifact sequences are
removed using the Cross_match software40, and an internally developed short
pattern finder. Low-quality regions of the read are identified using internally
developed software, which masks regions with a combined quality score of
o15. The longest high-quality region of each read is used as the EST. ESTs
shorter than 150 bp are removed from the data set. ESTs containing common
contaminants such as E. coli, common vectors and sequencing standards are
also removed from the data set. EST Clustering is performed ab initio, based on
alignments between each pair of trimmed, high-quality ESTs. Pair-wise EST
alignments are generated using the Malign software (Chapman J., personal
communication, JGI), a modified version of the Smith-Waterman algorithm41,42, which was developed at the JGI for use in whole-genome shotgun
assembly. ESTs sharing an alignment of at least 98% identity and 150 bp overlap
are assigned to the same cluster. These are relatively strict clustering cutoffs, and
are intended to avoid placing divergent members of gene families in the same
cluster. However, this could also have the effect of separating splice variants into
different clusters. Optionally, ESTs that do not share alignments are assigned to
the same cluster, if they are derived from the same cDNA clone. EST cluster
consensus sequences were generated by running the Phrap software40 on the
ESTs comprising each cluster. All alignments generated by malign are restricted
such that they will always extend to within a few bases of the ends of both ESTs.
Therefore, each cluster looks more like a ‘tiling path’ across the gene, which
matches well with the genome-based assumptions underlying the Phrap
algorithm. Additional improvements were made to the Phrap assemblies by
using the ‘forcelevel 4’ option, which decreases the chances of generating
multiple consensus sequences for a single cluster, where the consensus
sequences differ only by sequencing errors.
Genome assembly. The initial data set was derived from four whole-genome
shotgun (WGS) libraries: one with an insert size of 3 kb, two with insert sizes of
8 kb, and one with an insert size of 35 kb. The reads were screened for vector
using Cross_match, then trimmed for vector and quality. Reads shorter than
100 bases after trimming were then excluded. The data were assembled using
release 1.0.1b of Jazz, a WGS assembler developed at the JGI43. A word size of 14
was used for seeding alignments between reads. The unhashability threshold
was set to 50, preventing words present in more than 50 copies in the data set
from being used to seed alignments. A mismatch penalty of –30.0 was used,
which will tend to assemble together sequences that are more than B97%
identical. The genome size and sequence depth were initially estimated to be
16.5 MB and 9.3, respectively. The assembly contained 394 scaffolds, with
16.4 MB of sequence, of which 4.5% was gap. The scaffold N/L50 was 5/1.46
MB, whereas the contig N/L50 was 21/262 kb. The sequence depth derived
from the assembly was 8.77 ± 0.05.
Gap closure and finishing. To perform finishing, the P. stipitis whole genome
shotgun assembly was broken down into scaffold size pieces and each scaffold
piece reassembled with Phrap. These scaffold pieces were then finished using
our Phred/Phrap/Consed pipeline. Initially all low-quality regions and gaps
were targeted with computationally selected sequencing reactions completed
with 4:1 BigDye terminator/dGTP chemistry (Applied Biosystems). These
automated rounds included resequencing plasmid subclones and walking on
plasmid subclones or fosmids using custom primers. After completion of the
automated rounds, a trained finisher manually inspected each assembly.
Further reactions were than manually selected to complete the genome. These
reactions included additional resequencing reactions and custom primer walks
VOLUME 25
NUMBER 3
MARCH 2007
NATURE BIOTECHNOLOGY
© 2007 Nature Publishing Group http://www.nature.com/naturebiotechnology
ARTICLES
on plasmid subclones or fosmids. Again the reactions were completed using 4:1
BigDye terminator: dGTP chemistry. Smaller repeats in the sequence were
resolved by transposon-hopping 8-kb plasmid clones. Fosmid clones were
shotgun sequenced and finished to fill large gaps, resolve larger repeats or to
resolve chromosome duplications and extend into chromosome telomere
regions. After completion, each assembly was validated by an independent
quality assessment. This examination included a visual examination of subclone paired ends using Orchid (http://www-shgc.stanford.edu/informatics/
orchid.html), and visual inspection of high-quality discrepancies and all
remaining low-quality areas. All available EST resources were also placed on
the assembly to ensure completeness. All finished chromosomes are estimated
to have an error rate of less than 1 in 100,000 bp.
Gene prediction and annotation. The JGI Annotation Pipeline combines a
suite of gene prediction and annotation methods. Gene prediction methods
used for analysis of the P. stipitis genome include ab initio Fgenesh44,
homology-based Fgenesh+ (http://www.softberry.com/) and Genewise45, and
an EST-based method estExt (unpublished data). Predictions from each of the
methods were taken to produce ‘the best’ single gene model per every locus.
The best model was determined on the basis of similarity to GenBank proteins
and EST support. Every predicted gene was annotated using Double Affine
Smith-Waterman alignments (http://www.timelogic.com/) with Swissprot and
KEGG proteins. Protein domains were predicted using InterProScan46,47
against various domain libraries (Prints, Prosite, PFAM, ProDom, SMART).
Individual annotations have then been summarized according to Gene
Ontology48, KOGs18 and KEGG metabolic pathways49.
Phylogenetic tree reconstruction of sequenced fungal genomes. A multiple
sequence alignment of 94 single-copy genes present in 26 taxa was constructed
using the MUSCLE 3.52 program50, trimmed using Gblocks 0.91b and was
used as input for the maximum likelihood tree reconstruction program
PHYML (four rate categories, gamma + invariants, 100 bootstrap replicates)
resulting in a fully resolved tree with all but one node having bootstrap values
of 100. Figure 4 represents the portion of the tree describing relationships
between the genomes of interest for this analysis.
Comparative analysis of the six yeast genomes. Comparisons of the phylic
patterns of gene family distributions of P. stipitis and five hemi-ascomycete
yeasts (P. stipitis, S. cerevisiae, C. glabrata, K. lactis, D. hansenii and Y. lipolytica)
were done using the PhIGs orthology database19. The PhIGs resource generated
clusters of genes at each node on the evolutionary tree representing the
descendents from a single ancestral gene existing at that node. This allows
for the comparisons of the presence/absence patterns of gene families across the
six species avoiding confusion from paralogous genes. The set of 3,209 genes
determined to be orthologous from the PhIGs19 analysis were used to link
regions between the two genomes that represent orthologous chromosomal
segments with a minimum of four linking genes that are uninterrupted by
other orthology segments in either genome. In this analysis, gene families
specific to a single species are defined as those having a minimum of two
family members.
Expression analysis. To enable complete sampling of the expressed genes, we
generated four separate EST libraries by growing cells on glucose or xylose
under aerobic or oxygen-limited conditions. A set of 19,635 P. stipitis ESTs was
sequenced from the four libraries and clustered into 4,085 consensus sequences.
We mapped 94% (3,839) of the clusters to the genome and the numbers of hits
for each consensus cluster was used to estimate EST frequency under each
growth condition. An absolute majority of unplaced ESTs had problems with
the sequences so the data indicate completeness and accurateness of genome
assembly. Only 44% of the transcripts were represented by more than one EST
cluster-hit under any one of the four growth conditions. The cluster-hit
enumeration represents only a single biological sample for each of the four
conditions, so these observations must be interpreted with care and be limited
to the 200–400 most abundant gene models in which at least one transcript was
recovered under each of the four conditions. However, the relative abundances
of these ESTs under each of the four conditions provided a preliminary
expression analysis.
NATURE BIOTECHNOLOGY
VOLUME 25
NUMBER 3
MARCH 2007
Accession codes. P. stipitis genome assembly and annotations have been
deposited at DDBJ/EMBL/GenBank under the following accession numbers:,
chr 1.1 and chr 1.2, AAVQ00000000 and AAVQ01000000; chr_2, CP000496;
chr_3, CP000497; chr_4, CP000498; chr_5, CP000499; chr_6, CP000500; chr_7,
CP000501; chr_8, CP000502.
Note: Supplementary information is available on the Nature Biotechnology website.
ACKNOWLEDGMENTS
This work was performed under the auspices of the US Department of Energy’s
Office of Science, Biological and Environmental Research Program, and by the
University of California, Lawrence Livermore National Laboratory under
Contract No. W-7405-Eng-48; Lawrence Berkeley National Laboratory under
contract No. DE-AC02-05CH11231; Los Alamos National Laboratory under
contract No. W-7405-ENG-36; Stanford University under contract No. DEFC02-99ER62873, and by the US, Forest Service, Forest Products Laboratory.
The authors are grateful to C.P. Kurtzman of the USDA ARS Culture Collection
(NRRL) for providing the P. stipitis stock culture, to W. Huang, G. Werner and
his group of the JGI for engineering support of annotation, to A. Polyakov and
I. Dubchak of the JGI for VISTA analysis, to A. Darling for advice and support
in MAUVE analysis, W. R. Kenealy, T. A. Kuster and Mark Davis of the USDA
Forest Products Laboratory for carrying out continuous culture studies,
providing photomicrographs and analyzing fermentation products, Samuel
Pitluck and Kemin Zhou of JGI for assistance with the GenBank submission
and to James Cregg, and Lisbeth Olsson and Jennifer Headman Van Vleet for
critical readings of early drafts.
COMPETING INTERESTS STATEMENT
The authors declare competing financial interests (see the Nature Biotechnology
website for details).
Published online at http://www.nature.com/naturebiotechnology/
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions
1. Jeffries, T.W. Engineering yeasts for xylose metabolism. Curr. Opin. Biotechnol. 17,
320–326 (2006).
2. Saha, B.C., Dien, B.S. & Bothast, R.J. Fuel ethanol production from corn fiber - Current
status and technical prospects. Appl. Biochem. Biotechnol. 70–2, 115–125
(1998).
3. Kurtzman, C.P. Candida shehatae–genetic diversity and phylogenetic relationships with
other xylose-fermenting yeasts. Antonie Van Leeuwenhoek 57, 215–222 (1990).
4. Melake, T., Passoth, V.V. & Klinner, U. Characterization of the genetic system of the
xylose-fermenting yeast Pichia stipitis. Curr. Microbiol. 33, 237–242 (1996).
5. van Dijken, J.P., van den Bosch, E., Hermans, J.J., de Miranda, L.R. & Scheffers, W.A.
Alcoholic fermentation by ‘non-fermentative’ yeasts. Yeast 2, 123–127 (1986).
6. du Preez, J.C., van Driessel, B. & Prior, B.A. Ethanol tolerance of Pichia stipitis and
Candida shehatae strains in fed-batch cultures at controlled low dissolved-oxygen
levels. Appl. Microbiol. Biotechnol. 30, 53–58 (1989).
7. Hahn-Hägerdal, B. & Pamment, N. Microbial pentose metabolism. Appl. Biochem.
Biotechnol. 113–16, 1207–1209 (2004).
8. Nigam, J.N. Ethanol production from wheat straw hemicellulose hydrolysate by Pichia
stipitis. J. Biotechnol. 87, 17–27 (2001).
9. Nardi, J.B. et al. Communities of microbes that inhabit the changing hindgut landscape
of a subsocial beetle. Arth. Struct. Dev. 35, 57–68 (2006).
10. Suh, S.O., Marshall, C.J., McHugh, J.V. & Blackwell, M. Wood ingestion by passalid
beetles in the presence of xylose-fermenting gut yeasts. Mol. Ecol. 12, 3137–3145
(2003).
11. Lee, H., Biely, P., Latta, R.K., Barbosa, M.F.S. & Schneider, H. Utilization of xylan by
yeasts and its conversion to ethanol by Pichia stipitis strains. Appl. Environ. Microbiol.
52, 320–324 (1986).
12. Targonski, Z. Biotransformation of lignin-related aromatic-compounds by Pichia stipitis
Pignal. Zentralbl. Mikrobiol 147, 244–249 (1992).
13. Jin, Y.S., Laplaza, J.M. & Jeffries, T.W. Saccharomyces cerevisiae engineered for xylose
metabolism exhibits a respiratory response. Appl. Environ. Microbiol. 70, 6816–6825
(2004).
14. Passoth, V., Cohn, M., Schafer, B., Hahn-Hägerdal, B. & Klinner, U. Analysis of the
hypoxia-induced ADH2 promoter of the respiratory yeast Pichia stipitis reveals a new
mechanism for sensing of oxygen limitation in yeast. Yeast 20, 39–51 (2003).
15. Klinner, U., Fluthgraf, S., Freese, S. & Passoth, V. Aerobic induction of respirofermentative growth by decreasing oxygen tensions in the respiratory yeast Pichia
stipitis. Appl. Microbiol. Biotechnol. 67, 247–253 (2005).
16. Passoth, V., Hansen, M., Klinner, U. & Emeis, C.C. The electrophoretic banding pattern
of the chromosomes of Pichia stipitis and Candida shehatae. Curr. Genet. 22,
429–431 (1992).
17. Mayor, C. et al. VISTA: visualizing global DNA sequence alignments of arbitrary length.
Bioinformatics 16, 1046–1047 (2000).
325
© 2007 Nature Publishing Group http://www.nature.com/naturebiotechnology
ARTICLES
18. Koonin, E.V. et al. A comprehensive evolutionary classification of proteins encoded in
complete eukaryotic genomes. Genome Biol. 5 (2) Art. No. R7 (2004).
19. Dehal, P.S. & Boore, J.L. A phylogenomic gene cluster resource: the Phylogenetically
Inferred Groups (PhIGs) database. BMC Bioinformatics 7 Art. No. 201 (2006).
20. Dujon, B. et al. Genome evolution in yeasts. Nature 430, 35–44 (2004).
21. Fischer, G., Rocha, E.P.C., Brunet, F., Vergassola, M. & Dujon, B. Highly variable rates
of genome rearrangements between hemiascomycetous yeast Lineages. PLoS Genet. 2,
253–261 (2006).
22. Laplaza, J.M., Torres, B.R., Jin, Y.S. & Jeffries, T.W. Sh ble and Cre adapted for
functional genomics and metabolic engineering of Pichia stipitis. Enzyme Microb.
Technol. 38, 741–747 (2006).
23. Leandro, M.J., Goncalves, P. & Spencer-Martins, I. Two glucose/xylose transporter
genes from the yeast Candida intermedia: first molecular characterization of a yeast
xylose/H + symporter. Biochem. J. (2006).
24. Weierstall, T., Hollenberg, C.P. & Boles, E. Cloning and characterization of three genes
(SUT1–3) encoding glucose transporters of the yeast Pichia stipitis. Mol. Microbiol. 31,
871–883 (1999).
25. Chow, T.H., Sollitti, P. & Marmur, J. Structure of the multigene family of MAL loci in
Saccharomyces. Mol. Gen. Genet. 217, 60–69 (1989).
26. Jurka, J. et al. Repbase Update, a database of eukaryotic repetitive elements.
Cytogenet. Genome Res. 110, 462–467 (2005).
27. Neuveglise, C., Feldmann, H., Bon, E., Gaillardin, C. & Casaregola, S. Genomic
evolution of the long terminal repeat retrotransposons in hemiascomycetous yeasts.
Genome Res. 12, 930–943 (2002).
28. Daboussi, M.J. & Capy, P. Transposable elements in filamentous fungi. Annu. Rev.
Microbiol. 57, 275–299 (2003).
29. Parekh, S.R., Parekh, R.S. & Wayman, M. Fermentation of xylose and cellobiose by
Pichia stipitis and Brettanomyces clausenii. Appl. Biochem. Biotechnol. 18, 325–338
(1988).
30. Grotkjaer, T., Christakopoulos, P., Nielsen, J. & Olsson, L. Comparative metabolic
network analysis of two xylose fermenting recombinant Saccharomyces cerevisiae
strains. Metab. Eng. 7, 437–444 (2005).
31. Shi, N.Q., Cruz, J., Sherman, F. & Jeffries, T.W. SHAM-sensitive alternative respiration
in the xylose-metabolizing yeast Pichia stipitis. Yeast 19, 1203–1220 (2002).
32. Zakian, V.A. Structure, function, and replication of Saccharomyces cerevisiae telomeres. Annu. Rev. Genet. 30, 141–172 (1996).
33. Krogan, N.J. et al. COMPASS, a histone H3 (lysine 4) methyltransferase required for
telomeric silencing of gene expression. J. Biol. Chem. 277, 10753–10755 (2002).
326
34. Vidgren, V., Ruohonen, L. & Londesborough, J. Characterization and functional analysis
of the MAL and MPH loci for maltose utilization in some ale and lager yeast strains.
Appl. Environ. Microbiol. 71, 7846–7857 (2005).
35. Osborne, C.S. et al. Active genes dynamically colocalize to shared sites of ongoing
transcription. Nat. Genet. 36, 1065–1071 (2004).
36. Zhan, S., Horrocks, J. & Lukens, L.N. Islands of co-expressed neighbouring genes in
Arabidopsis thaliana suggest higher-order chromosome domains. Plant J. 45, 347–357
(2006).
37. Bartlett, O. et al. Specialized transcription factories. in Transcription vol. 73, 67–75,
(Portland Press Ltd., London, 2006).
38. Kaiser, C., Michaelis, S. & Mitchell, A.. Methods in Yeast Genetics (Cold Spring Harbor
Laboratory Press, Cold Spring Harbor, NY, 1994).
39. Burke, D., Dawson, D. & Stearns, T. Methods in Yeast Genetics: a Cold Spring Harbor
Laboratory Course Manual (Cold Spring Harbor Laboratory Press, Cold Spring Harbor,
N.Y., 2000).
40. Ewing, B. & Green, P. Base-calling of automated sequencer traces using phred. II. Error
probabilities. Genome Res. 8, 186–194 (1998).
41. Smith, T.F. & Waterman, M.S. Overlapping genes and information theory. J. Theor. Biol.
91, 379–380 (1981).
42. Smith, T.F. & Waterman, M.S. Identification of common molecular subsequences.
J. Mol. Biol. 147, 195–197 (1981).
43. Aparicio, S. et al. Whole-genome shotgun assembly and analysis of the genome of Fugu
rubripes. Science 297, 1301–1310 (2002).
44. Salamov, A.A. & Solovyev, V.V. Ab initio gene finding in Drosophila genomic DNA.
Genome Res. 10, 516–522 (2000).
45. Birney, E. & Durbin, R. Using GeneWise in the Drosophila annotation experiment.
Genome Res. 10, 547–548 (2000).
46. Zdobnov, E.M. & Apweiler, R. InterProScan–an integration platform for the signaturerecognition methods in InterPro. Bioinformatics 17, 847–848 (2001).
47. Mulder, N.J. et al. InterPro, progress and status in 2005. Nucleic Acids Res. 33,
D201–D205 (2005).
48. Ashburner, M. et al. Gene ontology: tool for the unification of biology. The Gene
Ontology Consortium. Nat. Genet. 25, 25–29 (2000).
49. Kanehisa, M., Goto, S., Kawashima, S., Okuno, Y. & Hattori, M. The KEGG
resource for deciphering the genome. Nucleic Acids Res. 32, D277–D280
(2004).
50. Edgar, R.C. MUSCLE: multiple sequence alignment with high accuracy and high
throughput. Nucleic Acids Res. 32, 1792–1797 (2004).
VOLUME 25
NUMBER 3
MARCH 2007
NATURE BIOTECHNOLOGY